Documente Academic
Documente Profesional
Documente Cultură
DOI: 10.5923/j.msse.20130202.02
Department of Electronics & Communication Engineering, National Institute of Technology Warangal, Warangal, 506004, A.P, India
Abstract In the deep sub-micron regime it is very crit ical to deal with the sub-threshold leakage currents. At different
levels of abstraction in chip design, this current constitutes more than fifty percent of the total leakage current. In this paper a
method for reducing sub-threshold leakage which uses Variable Gate Oxide Th ickness MOSFET as an alternate to the high
threshold voltage device is proposed. An inverter and a two -input NAND gate wh ich uses VGOT MOSFET are simu lated
using LT-SPICE. The simulat ion results show a betterment of 77% and 32.32% in sub-threshold leakage reduction for
inverter and two-input NAND gate respectively when compared with the sub-threshold leakage current of the circuits which
used no leakage reduction mechanism. The simu lation results also show an imp rovement of 50.64% and 15.15% leakage
reduction for inverter and t wo-input NAND gate respectively when co mpared with the conventional high threshold voltage
MOSFET circuits.
Keywords Variable Gate Oxide Thickness MOSFET, Sub-threshold Leakage Current, Oxide thickness, Threshold
Vo ltage, Weak Inversion Current
the oxide in the M OSFET has been varied uniformly[6] and MOSFET as a sleep transistor is shown in Figure.2. A sleep
non-uniformly[7], the device thus obtained is a Variable transistor cuts off the path from supply to ground when the
gate oxide thickness MOSFET. These devices when used as logic circu it is off. Thus the gates G1, G2 and G3 of VGOT
sleep transistors have led to significant reduction in the MOSFET (Figure.2) are operated such that the leakage
weak inversion current. current is min imized.
This paper uses the transistor level approach reported
in[6],[7] into circuits and the transistor model is used as
alternate to the conventional high threshold voltage
MOSFETs wh ich are more used for leakage minimizat ion.
This transistor model (exp lained in section 2) eliminates the
complex and costly fabrication process steps involved in the
fabrication of conventional high threshold voltage
MOSFETs. The fabricat ion of this transistor requires only
few process steps in addition to the process steps of a
normal M OSFET.
The structure of the VGOT M OSFET has been explained
in section.2, the usage of VGOT MOSFET in circu its such
as inverter and a two input NAND gate has been explained
in section.3. Section.4 elaborated the results of the inverter
and the two input NAND gate and in section.5 conclusions
were made and future scope of the work is suggested.
Figure 2. A Logic circuit with a VGOT MOSFET as sleep transistor
(instead of a high Vth transistor)
2. Variable Gate Oxide Thickness
(VGOT) MOSFET In most of the system designs it is observed that always
seventy percent of the circuits are found in standby mode
The structure of VGOT MOSFET shown in Figure.1 (either ON or OFF) which constitutes static current. If a
consists of stacked gates G1, G2 and G3. The input and the particular OFF circu it is to be made inactive it should be
transfer characteristics of the VGOT MOSFET reported made in such a way that its effect is not felt on the other
in[6],[7] are briefly described. As with the conventional parts of the circu it. In an inactive circuit the leakage current
MOSFET operation, when the MOSFET is subjected to the is most often found to be the weak inversion current[4].
drain voltage bias and the gate voltage bias such that Vgs < If a small circuit constitutes to the leakage, then in a
Vth weak channel inversion occurs. The operation is similar system design, each of the inactive parts constitute to more
to the normal M OSFET with gate G1. W ith Gate G2 the leakage current. Most often the components which are
sub-threshold current is lower to that of the sub-threshold inactive constitutes the overall leakage current of the system.
current of G1 and higher when compared with the In this situation a mechanism is needed which makes the
sub-threshold current using G3 as the gate terminal. This is weak inversion current very negligib le or reduced to some
the direct consequence of the rise in threshold voltage of the extent.
transistor as the oxide thickness increases.
3.1. Inverter
Figure.3 shows the implementation of an inverter with
VGOT MOSFET, M3 as the sleep transistor and the
transistors M1 and M2 form a normal inverter.
As the technology model for a VGOT MOSFET is not
modelled in LT-SPICE (Linear Technology-Simu lation
Program for Integrated Circu its Emphasis), the gate oxide
thickness of the transistor M3 is varied. The o xide
thicknesses of transistor M3 are chosen to be higher than
the oxide thicknesses of transistor M1 and M2. The oxide
thickness of transistor M3 is varied in accordance with the
values in the technology allowable range. An inverter is
implemented with 90n m technology and simulated in
Figure 1. Variable gate oxide thickness MOSFET
LT-SPICE.
The oxide thickness tox of the VGOT MOSFET is varied
3. Circuit Implementation in three iterations. In the first iterat ion the o xide thickness is
varied as if the gate G1 of VGOT MOSFET is biased.
An imp lementation of a logic circuit with VGOT Whenever the inverter is operated in active mode, the
26 Keerti Kumar. K et al.: Sub-threshold Leakage Current Reduction
Using Variable Gate O xide Thickness (VGOT) M OSFET
transistor M3 is turned on by biasing the gate G1 and the gate transistors the current flowing through the circuit is
inverter wo rks normally. In the second iteration tox is sub-threshold current, in which case the sleep transistor has
slightly increased as if gate G2 of the device has to be to be switched OFF and the logic circuit is disconnected
biased. In a situation if the functionality of the inverter is fro m the supply voltage.
not required for a specific operation, then the inverter can
be operated in the standby mode by biasing the gate G2
of the VGOT MOSFET appropriately. This directly depicts
the relationship between the threshold voltage and the gate
oxide thickness of MOSFET. The threshold voltage of the
MOSFET is a function of t ox, wh ich states that the threshold
voltage of the device is increased if o xide thickness is
increased. This makes the inverter isolated fro m the supply
voltage. So this offers a lo w weak inversion current, as the
gate of the VGOT M OSFET is still biased. In the third
iteration of tox is further increased as if gate G3 of VGOT
MOSFET is to be biased. If more reduction in weak
inversion current is expected then the gate G3 is biased, it
offers very less weak inversion current than the current
measured when the gate G2 is biased. The overall weak
inversion current measured for the circuit is lesser when the
terminals G2 and G3 are b iased.
4. Results
4.1. Inverter
As stated in section 3.1 the inverter is imp lemented in
90n m technology and simu lation results for sub-threshold
current with and without VGOT M OSFET are obtained.
Table.1 shows the value of weak inversion current and
correspondingly the current is p lotted against variation of
oxide thickness in Figure.5.
Figure 3. An Inverter with a VGOT MOSFET as sleep transistor The weak inversion current measured for the inverter
3.2. A Two-Input NAND Gate circuit without using the VGOT MOSFET is -90.002pA and
it is -192.381pA using a conventional high threshold
As another logic circu it with VGOT M OSFET as a sleep voltage transistor. Point A and point B in Figure.5
transistor, a two input NAND gate has been implemented corresponds to the weak inversion current without using
and the performance of VGOT M OSFET in reducing the VGOT M OSFET and using a conventional high threshold
sub-threshold current is studied. Figure.4 shows the voltage transistor respectively.
implementation of a two input NAND gate where the
Table 1. tox vs I sub for Inverter
transistors M1, M2, M 3 and M4 co mbine to form a CM OS
(Co mp lementary Metal Oxide Semi -conductor) NAND VGO T tox (nm) Isub (pA)
gate and the transistor M5 is the VGOT M OSFET acting as Without VGOT -90.002
a sleep transistor. As the technology used is 90nm
With High Vth -192.381
technology, the threshold voltage of the transistors in the
CMOS NA ND gate is 0.18V and the supply voltage is 1.5 -364.046
1.2V[8]. If the input voltages are higher than this threshold 1.6 -373.519
value the gate works normally provided transistor M5 is -389.353
1.75
turned ON. If the inputs are lesser than Vth of the NAND
M icroelectronics and Solid State Electronics 2013, 2(2): 24-28 27
architecture co-design for ultralow-power sub-threshold [5] Siva G Narendra, Anantha Chandrakasan, “Leakage in
operation," Very Large Scale Integration (VLSI) Systems, nanometer CM OS Technologies”, Springer 2006.
IEEE Transactions on, vol.13, no.11, pp.1213-1224, Nov.
2005. [6] K.Keerti Kumar, N.Bheema Rao, “Variable Gate O xide
Thickness MOSFET: A Device level solution sub-threshold
[2] Sung-M o Kang, Yusuf Leblebici, “CM OS Digital Integrated leakage current reduction”, Proc. of 2012 Intl conference on
Circuits analysis and design”, Third edition Tata Devices, Circuits and Systems.
M cGraw-Hill Edition 2003.
[7] K.Keerti Kumar, N.Bheema Rao, “Characterization of
[3] Jan M Rabaey, Anantha Chandrakasan, Borivoje Nikolic, Variable gate oxide thickness M OSFET with non uniform
“Digital Integrated Circuits A Design Perspective”, Second oxide thicknesses for sub-threshold leakage reduction”, Proc.
Edition, PHI 2003. of 2012 International conference on Solid state and
Integrated Circuit (ICSIC 2012).
[4] P.R.van der meer, A.van Stavaren and A.H.N van Roermund,
“Low power deep submicron CM OS logic sub threshold [8] BSIM 4.3.0 M OSFET Model – User’s M anual
current reduction”, Springer 2004.